A Review Of Cathy O’Neil’s “Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy”

(There are exceptions, of course, like the writings of Cory Doctorow.)

But in “Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy,”  Cathy O’Neil presents a concise case about the perils of Big Data through the examples she offers over decades of technological development, and this text will remain critically relevant in the years ahead. She addresses the pattern of fundamental flaws at the core of many of these systems and her cautionary remarks about increasing surveillance are perhaps the most pertinent points of the entire book.

Big_Bang_Data_exhibit_at_CCCB_17

Details of Big Bang Data exhibit at CCCB (Photo Credit: By Kippelboy (Own work) CC BY 3.0 (http://creativecommons.org/licenses/by/3.0), via Wikimedia Commons)

Each of the examples of Weapons of Math Destruction are characterized by intrinsic flaws. To identify these traits, she poses three questions to ask when examining any Big Data system:

First – Even if the participant is aware of being modeled, or what the model is used for, is the model opaque, or even invisible?

Second – Does the model work against the subject’s interest? In short, is it unfair? Does it damage or destroy lives?

And finally –  [has] the model the capacity to grow exponentially? As a statistician would put it, can it scale?

Throughout the book, O’Neil explores several examples of WMDs and their socio-economic consequences. The introduction presents how IMPACT scoring unfairly resulted in the termination of good teachers, and how WMDs routinely target the poor where they hurt the most. The first chapter outlines her work as a hedge fund quantitative analyst leading up to the collapse of the housing market. Predatory lending is a key example of a WMD. Next, she examines the feedback loop created by the U.S. News college ranking report, and the resulting skyrocketing of college tuition, as well as the predatory nature of enrollment marketing campaigns.

From there, she dives into UCLA’s PredPol system, designed to optimize police patrol of areas where crime is statistically most likely to occur, and how the system inherently targets impoverished neighborhoods, creating yet another feedback loop of increased incarceration. Another chapter outlines the negative consequences of automated resume analysis and job performance metrics, and how the “optimization” of work shifts negatively impacts the middle class and the working poor. The final chapters present similar flaws in data systems determining insurance rates and credit eligibility, as well as Big Data’s Orwellian impact on the political process of voter targeting.

While the world painted by these flawed systems may appear dour, the text is not without hope. Scott Galloway’s book, “The Four: The Hidden DNA of Amazon, Apple, Facebook and Google” painted the apocalyptic near-future where Apple, Google, Amazon, and Facebook serve as the four horsemen of the end times. But O’Neil’s concluding chapter offers a number of proposed solutions to implement checks and balances into these systems to prevent that sort of abuse and exploitation. O’Neil presents the informed insight of a woman in a field severely dominated by men, and her perspective of big data through the lens of moral conscience. She humanizes and personalizes the societal effect of these systems and makes the subject of algorithms engaging and impactful.

“Weapons of Math Destruction” effectively outlines the characteristic flaws shared by many Big Data systems throughout history, and presents practical measures to reign in these unchecked operations. It’s a sharp and relevant text for anyone interested in the way these technologies shape our culture.

Results of the Innerspace Labs’ Music Discovery Survey

The results are in for the Innerspace Labs Music Discovery Survey!  A huge thank-you to all who offered their input.

I created the survey out of a personal curiosity.  Sadly, I have very little contact with the general public outside of the few members of my digital publishing team at the office, and I wanted to know what impact the web has had on the ways listeners discover new sounds.

I suspected listeners utilized multiple media resources in their musical explorations and that certainly proved to be the case.  Contributors cited an average of 6.44 sources for new music data.  The majority of the music sources I offered as options for the survey were widely-used, save for rateyourmusic.com, music subreddits, Gnoosic, and Usenet groups which each accounted for fewer than 3% of users’ musical resources.  I found this particularly interesting as I visit RYM frequently as my primary ratings and review aggregator and find its information invaluable when researching artists and genres.

Survey Tablepsd

As expected, Youtube ranked as users’ most-used resource when sampling new sounds.  I was surprised, however, to find that radio, motion pictures, television, or other forms of mass media were the third-highest ranked information resource, right behind user’s own friends.  While I only see ~3 new films annually, and have no exposure to television or radio, it still appears that mass media is still a significant part of most people’s lives.

Spotify and other streaming services were the next-highest ranked source, accounting for 10% of listeners’ discoveries.  While they are not a viable source for non-commercial or analog-only recordings, they still offer an incredible convenience for quick-and-easy personalized radio stations and there is no shortage of articles proclaiming streaming the new standard for mass media.

Crate digging was another significant source, as were vinyl Facebook communities and private music forums.  I’m curious whether this is representative of the public at large or just for Innerspace readers, but it is exciting nonetheless.

I was similarly please by the results for music lit and other periodicals, which accounted for more than 5% of musical discovery.  While 5% doesn’t sound significant on the surface, bear in mind that users cited an average of 6-7 sources for new music, so I’m considering 5% a threshold for this survey.

Other sources of note are independent music blogs and local music performances, both of which were a delight to see still holding their own in the survey.  After attending the latest concert at my local university, I will certainly be visiting their music library for further research into works by their professors.

I’m also curious to see if torrenting will grow in popularity for general music research in the years ahead.  Personally, torrenting is a critical step in my music purchasing process.  I’ve yet to find a better system, whether for surveying the catalog of an artist or to compare various masters before investing my hard-earned cash.

I consider the survey a success as its certainly given me a better understanding of how users find new music.  Thanks once again to everyone who contributed!

How Innerspace Readers Discover New Music